Understanding the source of excitatory drive is critical in deciphering how a vertebrate locomotion central pattern generator (CPG) works. However, the interneurons providing this drive are poorly understood in most vertebrates. Xenopus tadpoles at hatching stage 37/38 are capable of swimming away when touched and the locomotion rhythm is generated and maintained by spinal cord and hindbrain neurons (Roberts, 2000). Recently, we have devised a way to make whole-cell recordings from tadpole spinal neurons and to label them with neurobiotin. We have obtained anatomical and physiological evidence that a group of spinal interneurons with ipsilateral descending axons produce excitatory drive during tadpole swimming. Paired recordings showed that these spinal excitatory interneurons (dINs), with anatomical features defined by neurobiotin filling, directly excited more caudal CPG neurons of all types including: (1) motorneurons, (2) commissural reciprocal inhibitory interneurons, (3) ascending interneurons providing inhibitory feedback to the skin sensory pathway and other CPG neurons, and (4) other dINs. Remarkably, we discovered that dINs corelease both acetylcholine and glutamate as transmitters (Li et al., 2004). This corelease occurred during both miniature EPSCs and unitary EPSCs produced by dINs. Apart from chemical synapses, dINs are also electrically coupled exclusively to other dINs. This coupling is weak and dye-coupling was not seen. Gap junction blockers Carbenoxolone, Heptanol and Flufenamic acid (FFA) could block the coupling but all had clear side effects. The electrical properties of dINs were examined. Importantly, when the current injection exceeded firing threshold, most dINs only fired one spike at the onset of current injection even when the current was increased up to 6 times above the threshold. dINs didn’t fire rebound spikes after negative current injection when the membrane potential was at rest. However, when the membrane potential was depolarised above firing threshold, reliable rebound firing was observed. During each swimming cycle, dINs fired one spike reliably and their spikes appeared earlier than other CPG neurons at similar longitudinal positions. Since dINs can directly excite all types of CPG neurons, this spike timing difference suggests that dINs are the source of the excitation that normally drives swimming. The column of dINs extends into the hindbrain area where some of them also have ascending axon branches. These ascending axons can form the base for excitatory feedback connections among these hindbrain interneurons. Such connections could underlie mechanisms to sustain swimming activity.